Mon. Not. R. Astron. Soc. 000. [TUTSI C20041 Printed 5 February 2008 (MN WF&t style file v2.2)
Tidal Capture of Stars by Intermediate- Mass Black Holes
in
o
o
(N
>
O
(N
>
(N
in
in
o
Oh
6
H. Baumgardt 1 *, C. Hopman 2 *, S. Portegies Zwart 3,4 * and J. Makino 5 *
1 Sternwarte, University of Bonn, Auf dem Hiigel 71, 53121 Bonn, Germany
2 Faculty of Physics, Weizraann Institute of Science, P.O. Box 26, Rehovot 76100, Israel
3 Astronomical Institute "Anton Pannekoek," , University of Amsterdam, Kruislaan 403, Netherlands
4 Section Computational Science, University of Amsterdam, Kruislaan 403, Netherlands
5 Department of Astronomy, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Accepted ????. Received ?????; in original form ?????
ABSTRACT
Recent X-ray observations and theoretical modelling have made it plausible that some
ultraluminous X-ray sources (ULX) are powered by intermediate-mass black holes
(IMBHs) . N-body simulations have also shown that runaway merging of stars in dense
star clusters is a way to form IMBHs. In the present paper we have performed TV-body
simulations of young clusters such as MGG-11 of M82 in which IMBHs form through
runaway merging. We took into account the effect of tidal heating of stars by the
IMBH to study the tidal capture and disruption of stars by IMBHs. Our results show
that the IMBHs have a high chance of capturing stars through tidal heating within a
few core relaxation times and we find that 1/3 of all runs contain a ULX within the age
limits of MGG-11, a result consistent with the fact that a ULX is found in this galaxy.
Our results strengthen the case for some ULX being powered by intermediate-mass
black holes.
Key words: globular clusters: general - black hole physics - stellar dynamics
;> , 1 INTRODUCTION
•i-H .
, Ultra-luminous X-ray sources (ULX) are point-like X-ray
5_l ■ sources with isotropic X-ray luminosities in excess of L =
10 40 erg/sec. Various theories have been proposed in the
literature concerning the nature of ULX, like stellar-mass
b lack ho le binaries with mild geometrical b eaming (King et
al l 20011 . Rappaport, Podsiadlowski & Pfahl 2005|i or high-
redshift quasars (Gutierrez & Lopez-Corredoir jj2005j) and
it is likely that ULX are not a hom ogeneous class of ob-
jects (Soria, Cropper & Motct i 20011) . Although luminous
X-ray sources have been observed in different kinds of envi-
ronments, sources with luminosities of L « 2 ■ 10 39 erg/sec
or brighter seem to be associated with young stellar pop-
ulations of star-for ming g alaxies (Irwin, Bregman & Athey
I 2004 Swartz et al l 20041) .
Since the Eddington luminosity of a star of mass M is
given by
M
with luminosities exceeding L > 10 40 erg s 1 could be
inte rmedi ate-mass black holes (IMBHs, see Miller & Col-
be H 20041 for a review). The starburst galaxy M82 for ex-
ample hosts a ULX with brightn ess in t he range L = (0.5 —
1.6) - 10 41 ergs" 1 (Matsumoto et a j 200 ll. Kaaret et al l 200 ill ,
corresponding to a black hole with mass 350 — 12OOM0 if
emitting photons at the Eddington luminosity. The case for
an IMBH in M82 is supported by a 54mHz quasi-periodic os-
cillation found in the X-ray flux (Strohmayer & Mushotzky
I 2003f) and also by t he sof t X-ray spectrum of this source
(Fiorito & Titarchul fiooll . Another argument supporting
the connection between ULX and IMBHs are cool thermal
emission components which have been found in the X-ray
spectra of some ULX and which can be fitted well with ac -
cretion disc models of IMBHs (Miller, Fabian & Mille l 20041) .
= 1.3 x lO^ergs"' —
Mr
(1)
where M is the mass of the accreting object, most low-
luminosity ULX are probably stellar-mass black holes. How-
ever, there is mounting evidence that the brightest ULX
* e-mail: holger@astro.uni-bonn.de (HB); clo-
vis.hopman@weizmann.ac.il (CH); spz@science.uva.nl (SPZ);
makino@ astron. s . u-tokyo .ac.jp ( J M )
Portegies Zwart et al. I (20041) have performed A^-body
simulations of the dense star cluster MGG-11 whose posi-
tion coincides with the ULX in M82. They found that run-
away merging of massive main-sequence stars in the centre
of this cluster leads to the formation of a massive star with
a mass of several 1000 Mq within a few Myrs if the initial
concentration of the cluster is larger than that of a King
Wo = 9.0 model. The connection between an IMBH and
the ULX was str engthened by Hopman, Portegies Zwart &
Alexander 1(20041) who showed through analytic estimates
that an IMBH with a mass of 1000 Mq residing in the cen-
tre of MGG-11 has a high probability of tidally capturing a
2 H. Baumgardt, C. Hopman and S. Portegies Zwart, J. Makino
passing main-sequence star or giant. The orbit of the cap-
tured star was found to circularise even when dynamical
perturbations by other cluster stars were taken into account
and the system entered a Roche-lobe overflow (RLOF) phase
once the captured star had sufficiently increased its radius
through stellar evolution.
P ortegies Zwart, Dewi & Maccarone If2004l) and Li (
2004) showed that IMBHs with mass-transferring compan-
ions near the end of their main-sequence lifetimes are able
to create X-ray luminosities in excess of L — 10 40 ergs _1
over several Myrs, giving them a high chance to be observed
as ULX. For sufficiently low-mass companions, the RLOF
mass transfer stage might start after the host cluster has
been dissolved by the tidal field of the parent galaxy, of-
fering a way to e xplain isolated ULX (Hopman, Portegies
Zwart & AlexandefjOO^).
In this paper we present the results of detailed iV-body
simulations of the formation and further evolution of an
IMBH in a cluster with characteristics similar to MG G-11 i n
the starburst Galaxy M82 (see Portegies Zwart et al l 2004) .
Our simulations included the effects of tidal heating and the
emission of gravitational waves of passing stars and we in-
vestigated the chances that MGG-f I produces a ULX within
the age limits determined by observations (7-12 Myrs).
This paper is organised as follows: In section 2 we re-
view the theory of tidal heating and section 3 gives analytic
estimates of tidal inspiral. Section 4 describes the imple-
mentation of tidal heating into our simulations. Section 5
discusses our results for MGG-11 and section 6 summarises
the paper and presents our conclusions.
is given by the tidal radius (see e.g. Kochanel TT992Tl :
2 TIDAL HEATING OF STARS NEAR A
MASSIVE BLACK HOLE
Stellar systems are usually well described as a system of
point particles interacting with Newtonian gravity. In a few
cases, however, when the stellar density is sufficiently high,
this approximation is no longer accurate and fails to describe
some of the important processes which occur in stellar clus-
ters. General relativistic effects for example become impor-
tant when stars approach each other at distances close to
their Schwarzschild radius, which can be relevant for com-
pact stars. Main sequence stars are much larger and close
encounters can lead to hydrodynamical interactions, with
stellar collisions as the extreme. This may lead to a runaway
merger in young and dense star clus ters (P ortegies Zwart et
al l 19991: Porte gies Zwart & McMilla Jiool Freitag, Giirkan
fc Rasi a 20051 : see also section 5).
Stars that pass each other at somewhat larger distances
may still affect each others orbits by energy dissipation in
a tidal interaction. Such an interaction may initiate tidal
modes on both stars. The energy invested to raise the tides
in the st ars is taken from their kinetic energies (Press &
Teukolsk l 19771) . Two initially unbound stars may so become
bound, or a bound eccentric binary can become bound more
tightly.
Stars can also have a tidal interaction with a black hole
(BH). The length-scale for tidal forces of an IMBH of mass
Mm on a star of mass M* and radius R* to become important
'M.\ 1/3
1 a/7>
(2)
We now recognise several distinct regimes: (1) The
pericenter distance is much larger than the tidal radius
(r p 3> r t ), in which case tidal effects are negligible. (2) The
distance of closest approach is smaller than the tidal radius
r v < rt- In this case the entire star may be disrupted or part
of its envelope may be stripped off to form a temporary ac-
cretion disk around the black hole. The gas accreting from
the star or the disc onto the black hole then causes a flare.
The duration of the flare depends on how violent the interac-
tion was and whether or not an accretion disc was formed:
for tidal disruptions around super-massi ve blac k holes the
flare might last at most a few years (Reei t 19 88). (3) In the
intermediate regime (r p > rt), the star is tidally deformed,
and orbital kinetic energy is transferred to the internal en-
ergy of the star. We call this the tidal capture regime.
For highly eccentric orbits (e > 0.9), the energy dissi-
pation per orbit can be parametrised by
21+2
Ti(v), (3)
R*
where
n :
M* + Mm
1/2
R,
3/2
(4)
Here Ti(jf) is the dimensionless tidal coupling function,
which depends on the stellar structure and is a strongly
decreasin g func tion of the pericenter r v (see e.g. Press &
Teukolsk l 19771) . The orbital energy at the tidal radius usu-
ally exceeds the binding energy of the star by several orders
of magnitude, and some fraction of the orbital energy is dis-
sipated at every new pericenter passage. As a result the star
becomes very hot and expands. Both effects make it more
luminous. The stellar luminosity can become much larger
than that achieved by nuclear burning, and is of order
Lt = — • (5)
Here P is the orbital period of the binary system.
It is unclear where a star stores the excess tidal energy.
In the two most extreme cases, the star can store the ex-
cess energy in the surface layers, or in the bulk of the stellar
material. The first case leads to high surface temperatures
without appreciable expansion of the star (McMillan, Der-
mott & Taar rl 19871) . whereas the latter causes the star to
remain cold but exp and dr amatically (Podsiadlowsk F"l996l) .
Alexander & Morri j 20031 dubbed the two possibilities as
"hot" and "cold" squeezars and argued that these tidally
excited stars may be observable in the center of the Milky-
Way Galaxy.
The long-term response of the star to tidal heating de-
termines whether the star survives the encounter. To be able
to survive the tidal circularisation near r t , a star on an ini-
tially wide orbit has to dissipate
Et
M. \ 2/3 GMl
R*
(6)
which generally is much larger than the binding energy of
the star. As a consequence, a star can only survive a strong
Tidal Capture of Stars by Intermediate- Mass Black Holes 3
tidal interaction if it cools efficiently. The most obvious way
this can be achieved is by radiation, i.e: a "hot squeezar".
The structure of a "hot squeezar" is unaffected by the
encounter, and as a result the cooling of the star is limited by
the Eddington luminosity. The limiting luminosity can now
be used to compute a lower limit on the pericenter for which
the star is able to survive an encounter. The further orbital
evolution can be described analytically and the time-scale
for com plete c ircularisation of the orbit is then (Alexander
& Morri i 2003ft :
t (r p ,a) =
2irMWGM.a
AE t (r p )
(7)
If this time-scale is sufficiently small, the orbit of the
star is not changed significantly by scattering with field stars
during inspiral (section l3.3H . In this case the star circularises
near the tidal radius of the IMBH. The subsequent evolution
of the IMBH with main-sequence star binary is discussed in
section 15.51 .
3 ANALYTICAL ESTIMATES OF TIDAL
INSPIRAL IN PRESENCE OF SCATTERING
If a star has a orbit in the tidal capture regime (r p > rt) af-
ter a first encounter with an IMBH, it experiences repeated
tidal interactions through which the orbit shrinks and and
becomes circularised. The time-scale for circularisation is
much larger than the orbital period P of the star (to ^> P).
If the orbit is perturbed by another star during the inspi-
ral process, the inspiraling star may either be scattered to a
wider orbit or into a tighter orbit. In the first case tidal heat-
ing will become less effective and the inspiral process slows
down or stops completely, making the star more vulnera-
ble for subsequent dynamical interactions. In the latter case
the inspiral process is either accelerated or, if the pericenter
distance becomes smaller than the tidal radius of the star,
the star is tidally disrupted by the IMBH and the inspiral
process ends.
3.1 Assumptions
We present an approach which is quite similar for tidal dis-
ruptions and tidal inspirals. To obtain analytical estimates
for the two processes, we make several assumptions that are
generally made for th e treat ment of tidal disruptions (e.g.
Lightm an and Shapir d 19771: Fran k & Ree J 1976t Cohn &
Kulsr u Jl97l Syer & Ulm d 1999| M agorr ian fc T remaine
I 19991: Miralda- Escud e & Goul d 20001: Fr eitat 2001al Alexan-
der & Hopmar pOOl Wan g fc M erri d 2004ft "and i nspiral
processes fe.g. Hils fc Bendel- 19951 : Sigurdsson & Rees 1997;
Ivano f200a Freita l 2001 bl 20031: Alexan der fc H opma £ 2003t
Hopman, Portegies Zwart & Alexande J 20041 : Hopman &
Alexande F2005Tl .
Our simple analytical model captures some of the phys-
ical mechanisms of the cluster and gives approximately the
correct estimates for the probability that a star is captured
by the IMBH. Since many of the assumptions listed here
have been made by numerous previous papers, it is of im-
portance to trace to what extend the analytical model agrees
with the simulations, which has less number of simplifying
assumptions.
Here we list our assumptions and discuss them.
(i) The stellar distribution function is well approximated
by a stellar cusp, i.e.
*(r) =
(3 - a)N a
(8)
where a is approximately constant withi n the radius of in-
fluence r a of the IMBH (Bahcall & Wol f 197fill 1977ft . and
N a is the number of stars within r a .
This assumption is shown t o be satisfied by earlier N-body
simulation s (Bau mgardt et al l 2004 J 2004b: Preto, Mcrritt
& Spurzem 2004), and also in our current simulations. The
slope of the cusp in our simulations is a ~ 1.5. A stellar cusp
with a « 1.5 has also been observed by s tar co unts near
the MBH in our Galactic Centre fAlexande f 1999ft . We will
assume a = 3/2 in the following analysis. This simplifies the
expressions somewhat, because in that case the relaxation
time is constant (see eq. 1121 1.
Baumgardt, Makino & Ebisuzaki: (20 04ah have shown that
for small-mass black holes, containing less than a few percent
of the total cluster mass, the radius of influence of the black
hole is given by
GM.
(9)
where a is the velocity dispersion in the cluster core.
(ii) Within r a the orbits of the stars are Keplerian to good
approximation. This is a good assumption for our purposes,
and it makes many of the expressions more transparent.
(iii) The relaxation time within the cusp is much larger
than the orbital time. Stars exchange energy and angular
momentum only through small angle two-body interactions.
(iv) Within the radius of influence the stellar population
can be approximated by a single mass population.
This assumption is made for simplicity, and much of the
discrepancy between the analytical model and the simula-
tions stems from this assumption. In a young stellar cluster
there is a wide range of masses, with a few stars which have
masses much larger than the mean stellar mass (M*). As
a result the most massive stars sink to the IMBH, and the
masses within the radius of influence vary strongly. This has
considerable consequences for the dynamical behaviour close
to the IMBH which the simple analytical model fails to de-
scribe. We will comment on this when we discuss the results
of our simulations.
(v) Stars reach the IMBH by diffusion of angular momen-
tum rather than energy.
This assumpt ion is a ccurate for a single mass distribution
(Bahcall & Wol f 1976ft . When a few very massive stars are
present in the cluster, however, dynamical friction operates
on a time-scale ia.f. ~ ((M+) / M*)t r , much shorter than the
relaxation time t r . The dynamical mechanism which drives
stars to inspiral orbits in a young stellar cluster therefore
typically operates in two phases. First there is an "energy
phase", during which a massive star loses energy to field
stars and sinks to the centre; this is a purely elastic phase
during which there is no tidal heating. Then follows an "an-
gular momentum phase", during which more interactions
with cluster stars change the angular momentum of the mas-
sive star, until it finally reaches an orbit with pericenter close
4 H. Baumgardt, C. Hopman and S. Portegies Zwart, J. Makino
enough to the tidal radius, and tidal heating becomes effi-
cient.
As a result of our simplifying assumptions, the following
analysis does not describe the radial distribution function
(DF) of the stars as a function of their mass. But once this
DF is given, it does provide a treatment of the interplay be-
tween scattering and dissipation which is in good agreement
with the simulations.
3.2 Tidal disruption
Tidal disruption of stars by an IMBH occurs when a star
on an eccentric orbit has angular momentum smaller than
the angular momentum at the "loss cone" (J; c ). For a main-
sequence star orbiting an IMBH, the loss-cone is defined as:
Jic = V2GM.r t .
(10)
The disruption rate and distribution of stars in an-
gular momentum were studied by Lightman & Shapiro (
1977). Here the average changes in momentum are much
smaller than Ji c (diffusive regime). For this case the time-
scale for changes of the p ericent er by (small angle) scattering
is (Alexander & Hopmar l 20031)
t p (r p , a) — —t r . (11)
a
Here t r is the relaxation time, which we can estimate with
P
tr=A A [ —
N(< r)
(12)
Here Aa is a constant which includes the Coulomb loga-
rithm, and 7V(< r) tx r s ~ a is the number of stars enclosed
within r. For a = 3/2 the relaxation time is independent of
distance from the IMBH.
The diffusive regime is then defined to be the region
where a < r cr i t ; r- cr it is the semi-major axis for which
-P(rcrit) = t p (r t , r cr it),
2?r
VGM.r t t,
2/5
(13)
The angular momentum of stars with a < r cr i t changes by
an amount smaller than J; c per orbit, while the change in
angular momentum per orbit is larger than Ji c if a > r cr it.
In the latter case the velocity distribution can be isotropic,
since it is not affected by the presence of the loss-cone. This
is not true for the diffusive regime.
The distribution function and merger rate in the diffu-
sive regime are computed by solving the appropriate Fokker-
Planck equation. The distribution function should become
isotropic for J 3> J; c and vanish at the loss-cone. The steady
state solution gives rise to a flow of stars through the loss-
cone, which is a consequence of the presence of a mass sink
at Ji c . This is reflected in the boundary condition that the
distribution function vanishes at the loss-cone.
The disruption rate in the diffusive regime is given by
daN iso (a) (r clit /r a ) 3/2 N a
i t r ln(J m / Jic) t r \n(J m / Jic)
(e.g. Lightman & Shaping 19771: Hopman & Alexande F^OO.
(14)
Here N(a)da oc a 2 ~ a da is the number of stars with semi-
major axes in the range (o, a + da). In eq. (1141 J m is to be
evaluated at 7- cr it.
Equation II H is interpreted as a slow diffusion of the
angular-momentum of the stars. The time-scale for the an-
gular momentum to change by a factor ~ J m = \J GM,a is
of the order of the relaxation time t r . In a time t r a signifi-
cant fraction of the stars within 7- cr it are therefore disrupted
by the IMBH. The disruption rate is suppressed by a log-
arithmic factor which reflects the presence of the loss-cone.
This leads to a dilution of stars with small angular momenta
in the diffusive regime, as can be seen from the s olution of
the Fokker-Planck equation (Hopman & Alexander 2005).
For the tidal disruption rate, there is also a contribution
from stars for which the change in angular momentum per
orbit is larger than J; c (i.e., the regime a > r cr i t ). In that
case the disruption rate T(< a) for stars with semi-major
axis smaller than a is estimated to be given by the fraction
of stars in the loss-cone, divided by the period,
I* ick (< a)
N(< a)r t
aP(a)
(15)
For tidal capture (see below) there is no such con-
tribution, and we do not further consider stars in this
"kick" reg ime h ere; for a discussio n see, e.g., Lightman
fc Sh apiro [7l977h : Cohn & K ulsrud ffi97gh : Syer & Ulmer
Ifl999h : Magorrian & Tremaine f?1999T) .
3.3 Tidal capture
Due to relaxation in angular momentum, any star will even-
tually be accreted by the IMBH, either because of direct
tidal disruption, or due to tidal heating and inspiral. If
the number of stars in (a, a + da) which spiral in is given
by Ni(a)da, and the number of stars which are directly
disrupted by Nd(a)da, then the probability that the star
reaches the IMBH by inspiral is given by
S(a) =
Ni(a)
N t (a) + N d (a)
(16)
Hopman & Alexander If2005h performed Monte Carlo
simulations to show that stars with high energies (small
semi-major axes) are unlikely to enter the IMBH without
first spiralling in, while stars with low energies (large semi-
major axes) are tidally disrupted without experiencing a
tidal inspiral phase.
The rate of inspiral is then given by
r, =
1 daN(a)S(a)
t r \n(Jm/Jlc)
(17)
Here the function S(a) to accounts for the fact that capture
without premature disruption is improbable for a > a c .
In order for the star to survive, tidal heating should hap-
pen at a sub-Eddington rate (see previous section), yielding
a minimal value for the pericenter r p , m j n ~ 2r t . Inspiral
without premature scattering is only feasible provided that
Tidal Capture of Stars by Intermediate- Mass Black Holes 5
to(r p ,a) < t p (r p ,a). As a consequence stars can only spiral
in if initially their semi-major axis is smaller than
SAEt (^*p,min)^*p, miliar
1 2/3
(18)
Note that to > P and thus a max < r cl it, which implies that
for tidal capture the relevant regime is the diffusive regime,
and that the tidal capture rate is smaller than the rate for
tidal disruption, in spite of the fact that tidal heating is
efficient for r p < 3rt, while tidal disruption happens only
for r p <r t .
The function S(a) may be approximated by a Heavi-
side step- functio n 8(a — a c ), where a c > a max (Hopman &
Alexande ilooil . This approximate behaviour is confirmed
by our simulations (see JSJ. The rate at which stars undergo
inspiral due to tidal heating and survive is then given by
r,
: daTV(q)
t r ln( J m j Jl c
(amax/r a ) 3/2 jV a
t r ln(J m /Ji c )
(19)
(20)
Here the radial stellar distribution function is given by Eq.|H]
with a — 3/2.
Comparing Eas. l20l and ll4l implies that the rate at which
stars in the diffusive regime are tidally disrupted is larger
than the tidal capture rate by
r,
^"crit
3/2
(21)
The actual tidal disruption rate is even higher, because
stars with semi-major axes a > r cr i t can also be disrupted.
4 DESCRIPTION OF THE TV-BODY RUNS
We simulated the evolution of MGG-11 through TV-body
simulations of star clusters containing TV = 131,072 (128K)
stars u sing A arseth's collisional TV-body code NBODY4
fAarset fl999T) on the G RAPE6 computers of Tokyo Univer-
sity (M akino et all_2003). Following Portegies Zwart et al. (
2004), our clusters are initially King models with Wo = 9
and half-mass radius — 1.3 pc. The initial mass spectrum
of the cluster stars was a Salpeter mass-function between
I.OMq ^ m ^ 1OOM0. These cluster models have a pro-
jected half-mass radius, mass-to-light ratio and total cluster
mass after 12 Myrs that is consistent w ith the properties of
MGG-11 as observed by McCrady et al. lf2003l) . The clusters
are also concentrated enough to form IMBHs via runaway
merging of st ars. St ellar evolution was modelled according to
Hurley et al. [(200 J) and individual runs were calculated for
T = 12 Myrs. Table 1 gives an overview of the simulations
performed.
Stars were merged if their separation became smaller
than the sum of their radii and the mass of the merger prod-
uct was set equal to the sum of the masses of the stars, i.e.
we assumed no disruption of stars and no mass loss during
a collision. Since typical velocities of stars in our clusters
are much smaller than the escape velocities from stellar sur-
faces, this should be a valid assumption. The mass of tidally
disrupted stars was added to the mass of the central black
hole.
In total we performed 12 runs. In six of the runs we
used a relation for the radius of the runaway star d erived
from a fit to the results of Ishii, Ueno & Kato lfl999l) given
by
1 R * 1 K i M *
log i^ = L51og Me
1.85.
(22)
for stars with 120M Q < Af* < 1000M©. Radii of more
massive stars were set equal to the size of 1O 3 M0 stars.
These runs created fairly massive IMBHs with masses be-
tween 3OOOM0 < m < 4000Mq. In another set of runs, we
used t he mass-radius relation found by Bond, Arnett & Carr
Ifl984h :
log-gl = 0.47 log -^ + 0.20.
ft© M e
(23)
While derived for stars with masses larger than 10 4 Mq,
this relation gives radii in good agreement with the ZAMS
radii from Hurley et al. for stars with masses around 10 2 Mq.
These runs lead to runaway stars with masses around
1000-2000M Q , roughly equal to the estimated mass of M82
ULX-1. We did not apply stellar mass- loss for the runaway
stars and transformed them into IMBHs at T = 3 Myrs.
4.1 Implementation of tidal heating into the
TV-body simulations
We have implemented orbital energy loss by tidal interac-
tions between a star and the IMBH in the TV-body simula-
tions using the prescription of Portegies Zwart & Meinen (
I 19931 . henceforth PZM). They present fitting formula to the
tidal energy loss as a function of the masses, radii and poly-
tropic indices of passing stars. The dimensionless functions
Ti(rf) in eq. @ for the tidal energy were obtai ned fr om a
polynomia l fit to calculations by Lee & Ostrikcr (1986) and
Ray et al. Ifl987j) for stars of polytropic indices n = 1.5,2
and 3. PZM found that the / = 2 and 1 = 3 terms give a
99% contribution to the dissipated energy and therefore did
not take higher harmonics into account.
In our implementation, we assumed that main sequence
stars with masses m < O.4M0 and giants have polytropic
index n = 1.5, while stars with mass O.4M0 < m < 1.25M0
have polytropic index n — 2. More massive main-sequence
stars were assumed to have n — 3. It was not necessary to
apply tidal heating to white dwarfs or neutron stars since
these stars were not present in the runs because the studied
clusters are too young.
Eq. 3 is only valid if the orbital eccentricity is close to 1.
For orbits of smaller eccentricity the amount of tidal heating
is smaller since the star feels the influence of the IMBH also
at apocenter. In order to reflect this we multiply the heating
energy AEt by the eccentricity e of the orbit. In our current
implementation, the total angular momentum is conserved
and an unperturbed IMBH-star system will always circu-
larise at twice the initial pericenter distance r P i.
Due to the structure of NBODY4, tidal heating had to
be included in three different places in the code. In the main
integrator, we checked the distance of each particle to its
nearest neighbour each time it was advanced in time. This
is facilitated by the GRAPE hardware which provides the
6 H. Baumgardt, C. Hopman and S. Portegies Zwart, J. Makino
Table 1. Details of the performed N-body runs. Shown are the mass-radius relation assumed for the runaway stars (Ishii et al. or Bond
et al.), the mass M. j of the IMBH at time of its formation (T=3 Myrs), the mass M, f of the IMBH at the end of the run (T=12 Myrs)
and the number Njjis of stars tidally disrupted by the IMBH. Final columns give orbital parameters for tidal heating events: Starting
time Ti nsp of the inspiral, duration of inspiral ATj nsp until the orbit is circularised, mass M* and radius ij* of the star at time of inspiral,
tidal radius Rt of the star, initial pericentre distance of the orbit r p j, ratio of pericentre distance to the tidal radius of the star r p ;/r t ,
semi-major axis a and eccentricity e of the initial orbit and semi-major axis r Pi f after the orbit circularised.
Run M-R
M B h i
M BH f
R*
Rt
r p i 1 Rt
a
e
pj
rel.
[M Q ]
[Mq]
[Myr]
[yrs]
[Mq]
[Rq]
[Rq]
[Rq]
[Rq]
[Rq]
i nviivyy
3572.4
3667.5
29
3.16
1.1
10 2
6.12
3.0
26.2
26.5
1.01
13375
0.998
53.0
Q O/l
o.y4
1.2
10 5
l.O 1
1 A
1.4
1 n q
tro A
Do. 4
Q (11
o.Uo
i a no
14Uo
u.yoy
nn o
yu.z
10.12
1.7
10 3
2.02
1.6
20.4
21.8
1.07
34149
0.999
58.9
z iiviivyy
4076.3
4188.9
27
3.05
2.5
10 2
1.49
1.4
20.8
35.1
1.69
33410
0.999
42.9
8.92
1.9
10 2
1.77
1.5
21.1
36.6
1.73
Iboo ( y
0.999
38.8
o nviivyy
2/86.2
QOQO *7
6161. 1
31
3.13
5.7
10°
64.97
AAA
161.1
383.7
2.38
14695
0.974
728.2
Q A 1
0.41
7 ^
I .0
1U
en A O
0U.4Z
oU.O
1 fin q
i6y.o
1.4U
OODOO
u.yy (
QQft A
OOD.4
4.80
f, 7 ■
U. (
1 n3
25.61
1961.8
1 0082 9
16172.1
1.60
- ) SL\J\) _L \J
0.949
6.13
1 ^
l.O '
17.64
8.3
48.3
113.7
2.36
22883
0.995
120.3
10.03
Z.O '
i n 5
1.35
1.3
18.3
36.2
1.98
76953
0.999
69.2
4 IMK99
3325.3
3783.8
40
3.13
3.7
10 2
48.84
22.5
96.2
149.1
1.55
100084
0.998
196.9
3.42
9 n
z.u ■
49.75
28.9
124.4
185.5
1.49
31597
0.994
370.9
3.77
4:.U
39.68
20.5
95.8
130.2
1.36
21103
0.994
252.4
10.91
9 n
z.u
1
3.01
2.0
22.4
23.8
1.06
25552
0.999
47.4
5 IMK99
3190.3
3464.0
18
3.14
9 n
Z.U '
61.72
37.5
144.7
485.1
3.35
20418
0.976
447.8
3.55
o.o
10 3
1.43
1.4
19.2
24.6
1.28
117399
0.999
51.0
5.28
6.3 •
10 1
2.32
1.7
20.3
23.1
1.14
43542
0.999
42.9
7.74
9 7
Z. 1
1 n 4
21.60
988.0
5499.1
10847.7
1.97
372902
0.971
14373.1
11.69
3.6 •
10 4
15.69
682.7
4249.3
5530.9
1.30
363884
0.998
8503.8
6 IMK99
3603.7
4230.2
39
3.15
7 Q
i n 4
65.53
47.3
185.5
688.0
3.71
9642
0.929
623.5
3.53
3.0
10 3
46.70
27.5
122.1
565.4
4.63
215807
0.997
248.5
3.59
9 Q
z.y *
1 n3
29.15
11.8
61.8
154.7
2.50
133332
0.999
199.1
4.45
4.9
10 3
37.05
31.2
152.1
192.8
1.27
82735
0.998
385.4
4.98
9 ^
z.o
i n3
1U
25.41
12.7
70.9
120.8
1.70
49705
0.998
222.3
8.01
9 n
z.u
1U
9.86
1489.9
11479.4
14360.4
1.25
655428
0.978
22904.6
1 d A no/i
1 OQO 1
iyou. i
zoy ( .o
Zo
Q f\A
O.U4
1.5
10 4
4o.o0
OK 1
Qfl 1
A A Q O
4.y /
lyouo
n (177
u.y ( (
ooo o
3.84
6.8 •
10 4
/in T A
40. 1 4
42.5
166.0
Q A O C
348.0
2.10
30596
0.989
505.2
o. / y
6.6
10 3
z.Uo
1 R
l.D
1 1 . {
or; o
zo.z
1 A O
1.4z
40 (DO
n non
u.yyy
01. I
Q D Apo,!
ZU04.0
Zooo.l
O.U4
3.2
10 2
fio nA
Q/t fl
o4.y
116. U
1 Oft Q
i nn
i.uy
io * syo
u.yyy
O/l Q O
Z4o.Z
3.40
1.2 •
10 3
43.49
20.6
79.7
385.2
4.83
56733
0.993
168.7
y r5At^o4
1 1 A 1 O
114 / .Z
lo /y.o
Q /I O
o.4Z
2.2 •
10 4
a a nn
44. UU
z / .z
oo.z
1 m /l
iyi.4
Z.ZO
zlln /
n nm
u.yy 1
ZOO. (
4.11
3.2 •
10 2
37.93
32.8
112.7
193.8
1.72
251
0.229
238.1
4.95
7.4-
10 3
27.91
17.3
67.7
177.1
2.62
27882
0.994
195.4
6.38
4.4-
10 5
12.45
5.5
28.5
93.0
3.27
33949
0.997
130.2
11.82
1.8 ■
10 5
12.81
843.1
4328.9
11232.9
2.59
294365
0.962
12043.5
10 BAC84
1655.3
2009.4
31
3.16
1.1 •
10 4
50.05
55.3
185.0
250.7
1.35
70838
0.996
368.1
4.42
1.8 •
10 3
35.02
31.9
121.3
239.7
1.98
22510
0.989
264.7
4.93
7.9-
10 3
31.89
32.4
128.2
293.7
2.29
3538
0.917
284.5
5.12
6.2 •
10 3
28.84
1738.4
7161.1
22012.6
3.07
190684
0.885
14997.0
5.83
1.5 •
10 5
24.35
15.6
68.5
195.8
2.86
45854
0.996
255.2
6.33
2.6 ■
10 4
13.34
5.9
31.9
68.3
2.14
145350
0.999
105.7
9.90
8.4-
10 4
16.94
15.1
75.8
179.5
2.37
44328
0.996
256.5
11.13
1.4-
10 5
16.13
708.0
3629.3
5552.3
1.53
80890
0.931
9856.8
11 BAC84
1207.3
1565.3
19
5.36
7.4-
10 5
30.87
27.8
104.3
189.4
1.82
49832
0.996
417.8
6.00
6.8 ■
10 4
24.54
1357.5
5524.5
22582.5
4.09
104937
0.785
16732.5
10.06
5.1 ■
10 2
16.64
14.1
65.9
81.3
1.23
41275
0.998
139.7
12 BAC84
1700.8
2287.7
20
3.30
3.2 •
10 2
27.27
10.5
44.2
92.4
2.09
5641
0.984
90.0
8.32
1.2 •
10 4
11.89
1385.5
8154.8
9569.0
1.17
212502
0.955
16101.8
10.12
2.5 •
10 4
16.99
58.9
309.9
924.1
2.98
49761
0.981
1629.2
11.37
1.1 •
10 3
2.33
1.7
17.9
43.7
2.44
89189
0.999
44.2
index of the nearest neighbour for each particle at almost
no extra computational cost. The closest encounter distance
was calculated once the relative velocity v re i — ^ij^ij
between two particles switched sign from negative to pos-
itive and we applied tidal heating to all stars that passed
within 5r t of a massive black hole by decreasing the kinetic
energy of the combined star-black hole system by an amount
AE t . Energy was extracted by changing the velocity and po-
sition of the black hole and the star when the system was
at pericenter. The energy was changed according to eq. 3
such that the total linear and angular momentum remains
constant.
Tidal Capture of Stars by Intermediate- Mass Black Holes 7
r [RJ
10 6 10 7
60
0.0001
100
Figure 1. Density profile of run 6 at T = 3.15 Myrs, just before
an inspiral event happens. Between 10 pc < r < 0.05 pc, the
density profile of the cluster follows a power-law cusp with a =
1.5. The innermost data point is due to one star which experiences
an inspiral at T = 3.15 Myrs.
Since in NBODY4 the motion of close particle pairs
is foll owed by KS regularisation (Kustaanheimo & Stiefel
I 19651) . we also had to add tidal heating to the motion of
KS binaries. For unperturbed pairs this was done each time
their motion was integrated by changing the semi-major axis
and eccentricity of the binary according to the accumulated
amount of tidal energy. For perturbed binaries, tidal heat-
ing was applied after each pericenter passage since at this
point the perturbation of the motion of close neighbours is
smallest and it is easiest to correct the total energy of the
cluster.
Energy dissipation by grav itational radiation was imple-
mented following Peters (1964) formulae in a similar fashion
as the implementation of tidal heating.
5 RESULTS
5.1 Central density and mass segregation
We begin by discussing the density profiles of the star clus-
ters after an IMBH has formed in the centre. Fig. 1 depicts
the density profile of run 6 from Table 1 at T = 3.15 Myrs
just before a tidal capture event occurs. The cluster centre
is assumed to coincide with the position of the IMBH. It
can be seen that the cluster follows a power-law mass den-
sity profile p ~ r~ a with slope a — 1.50 inside r = 0.05
pc. This is close to the va lue of a = 1.55 found by Baum-
gardt, Makino & Ebisuzaki[?2bo4b) for IMBHs in multi-mass
clusters. Inspection of the calculations shows that a cusp is
already present by the time the runaway star has collapsed
40 -
20
— i — 1 — 1 — 1 — i — 1 — 1 — 1 — r
Average stellar mass
Maximum mass
Innermost star
Cusp stars
8
T [Myrs]
10
12
Figure 2. Average and maximum masses of stars in the cluster
as a function of time. The masses of all stars decrease as a result
of stellar evolution. Due to mass segregation, stars in the cusp
around the IMBH are more massive than the average cluster stars
and the star closest to the IMBH is always one of the most massive
stars in the cluster.
to a black hole (at about 3 Myrs), and as a result already
influences the late stages of runaway merging. One reason
for the quick formation of cusps is the large initial mass
spectrum of our models which reduces the relaxation time.
Fig. 1 also justifies our assumptions about the presence of
a cusp around the IMBH in §3. Due to the finite number of
cluster stars, the central cusp runs out of stars at r ~ 10~ 3
pc, corresponding to r = 10 5 Rq. In the present example,
one star was pushed to r = 10~ 4 pc due to dynamical inter-
actions between the inner cusp stars. This star experiences
an inspiral event at T = 3.15 Myrs.
Fig. |21 shows average and maximum masses of stars in
the cluster and near the IMBH over the course of a simu-
lation. Values shown are averaged over several simulations.
Masses of high-mass stars decrease as a result of stellar evo-
lution since stars lose mass in winds and the most massive
stars are constantly removed and turned into compact rem-
nants. As a result, the mass of the most massive cluster star,
which is near 80 Mq at T = 3 Myrs, drops to 15 Mq at the
end of the simulation. Due to the large number of low-mass
stars, the average mass of cluster stars stays nearly constant
within the first 10 Myrs at (m) ~ 3M®. Due to mass seg-
regation, the mean mass of stars in the cusp is higher than
the mean mass of all cluster stars. Also within the cusp,
massive stars sink close to the IMBH so the stars near the
IMBH have masses significantly higher than the mean mass.
This shows that it is important to take mass segregation into
account when determing the inspiral rates.
5.2 Tidal capture
Several trends can be deduced from Table 1:
8 H. Baumgardt, C. Hopman and S. Portegies Zwart, J. Makino
• Successful inspiral events occur in all runs with on av-
erage ~ 4 inspiral events per run during the first 9 Myrs
after IMBH formation. Hence there is a high chance that an
IMBH in a young star cluster like MGG-11 will undergo a
RLOF phase at least once during the lifetime of the cluster.
• Since the masses of stars in the inner cusp are relatively
large (see Fig. |2J , most inspiral events also involve massive
stars. The average mass of captured stars is (m) w 25Mq,
which exceeds the average mass of stars in the cusp. This
is caused by the larger cross-section for a tidal interaction
with the IMBH of the massive stars.
• In agreement with the theoretical estimates of §3, suc-
cessful tidal inspiral occurs only for stars in tightly bound
orbits around the IMBH with semi-major axis a < 10 5 -Rq,
the exception being giant stars with much larger stellar radii.
This corresponds roughly to the inner end of the cusp. Stars
on orbits with larger semi-major also experience tidal inter-
actions with the IMBH (see Fig. |7J , but are scattered away
by cusp stars before their orbits can circularise.
• Although the cross-section for tidal heating is larger
than the cross-section for tidal disruption, it is much more
likely to tidally disrupt stars, than to bring a star into a close
orbit around the IMBH via tidal heating. Averaged over all
runs, we find that the ratio of tidal disruptions to inspirals
is N d /Ni w 7.
• Successful inspiral events occur predominantly early on
in the cluster evolution: Between 5 < T < 12 Myrs only
24 inspiral events occur compared to 26 events between
3 < T < 5 Myrs. The decay in the number of tidal inspi-
ral events is mainly caused by the expansion of the cluster,
which is driven by mass loss from stellar evolution. In addi-
tion, the most massive cluster stars leave the main sequence
and increase their radii at T w 3.3 Myrs, which explains the
large number of inspirals around this time involving stars
with masses m > 25M©.
• The final pericenter distances after the orbit circu-
larised, are within a factor of three of the tidal radii since
the timescale for tidal circularisation is small only for small
r p /r t .
As an example for an inspiral event, Fig. [3] depicts the
evolution of semi-major axis a and pericenter distance r p of
a captured m — 17.6 Mq star with time from one of our
runs (run 3). The time when tidal heating shrinks the orbit
is shown by a solid line. Between 5.3 and 6.1 Myrs the star
segregates towards the IMBH as a result of its higher than
average mass and the semi-major axis of the orbit shrinks
to a few 10 4 Rq. Tidal heating occurs at T = 6.13 Myrs,
when the pericenter distance is scattered to a distance only
slightly larger than the tidal radius. During the inspiral the
semi-major axis decreases from more than 10 4 R© to about
120 Rq, and the orbit circularises at r p f/Rt = 2.5. After
the tidal circularisation, the semi-major axis remains un-
changed except for perturbations due to passing stars, which
increase the eccentricity. Upon each time a small eccentric-
ity is induced, tidal effects will directly start to circularise
the orbit again, and so result in a slight decay of the semi-
major axis. A strong orbital perturbation causes the star
to fill its Roche-lobe at T ~ 10.2 Myr and it is destroyed
by the IMBH. Our present TV-body code does not allow to
treat RLOF phases, a star whose stellar radius expands or
whose orbital radius shrinks to the point where the stellar
10 6
s 10 5
1000
100
' ' ' 1 ' ' ' ' 1 ' ;
Pericenter distance r p .
— Semi major axis a "1
— Tidal radius
— \
V
•
— Tidal heating phase
«
, , , 1 , , , , 1 , -
_l I I I I I I I I I L_
5.5 6 6.5
T [Myrs]
Figure 3. Semi-major axis and pericenter distance of a tidally
heated star before and after inspiral due to tidal heating. The
main tidal heating event happens at T = 6.13 Myrs at which
time the star circularises. Later, the semi-major axis changes due
to external perturbations while the tidal radius increases as a
result of stellar evolution and IMBH growth.
radius exceeds the tidal radius will simply be disrupted. We
will present a discussion about the further evolution of such
systems based on simple binary-stellar evolution estimates
in chapter 5.5.
5.3 Super-Eddington heating rates
The inspiral process may lead to complete tidal circulari-
sation, but also to the destruction of the star. The latter
happens if 1) the star passes the black hole's event horizon
or 2) if the rate of energy dissipation in the tidal encounter
exceeds the Eddington luminosity. We will now estimate the
latter fraction. For each inspiral event we calculated the
rate at which energy was pumped into the star due to tidal
heating when the eccentricity of the orbit was e = 0.9. At
this time the amount of energy dissipation per time AE/P
reaches a maximum since a and P decrease with e while for
smaller eccentricities the stars feel the influence of the black
hole also at apocenter and AE decreases.
Fig. [I] compares the energy dissipation rates at e =
0.9 with the Eddington luminosities of the stars. Super-
Eddington heating rates occur in roughly half the cases,
but since Fig. [I] shows the maximum heating rates during
the inspiral which occur only for a few orbits, cases with
AEi„s P > LEdd do not necessarily end up in the destruc-
tion of the star. Stars might for example lose their outer
envelopes and shrink, thereby becoming less susceptible to
tidal perturbations, so that the central parts of the stars
survive the inspiral. In addition, the extra energy pumped
into the star must be larger than the potential energy of the
star in order to disrupt it, so the super-Eddington heating
rates have to be maintained over many orbits. In order to
Tidal Capture of Stars by Intermediate- Mass Black Holes 9
Insp'
Figure 4. Energy dissipation rates AEi nap when the orbital ec-
centricities have reached e = 0.9 compared with the Eddington
luminosities L^dd of the stars. Nearly half of all stars experi-
ence sub- Eddington heating rates during the inspiral process and
survive the inspiral. Moderately super-Eddington stars might also
survive the inspiral since the super-Eddington heating rates occur
for only short intervals and the extra energy pumped into the star
is not sufficient to disperse it completely. If the amount of tidal
heating is too large, the stars expand and are tidally disrupted
by the IMBH.
answer the question which stars survive the inspiral, more
detailed simulations treating the response of the star to the
tidal heating would be required. These are beyond the scope
of the present paper.
5.4 Comparison with analytic estimates
We compare the numerical results with the analytical es-
timates from for the following rather typical example:
M. = 3 x 1O 3 M , M* = 30M Q , = WR Q , t rc = 3 Myrs,
r a = 0.05 pc, N a = 300, and r p /r t = 2.3. With eqs.
(14), (19) and (21) we then find that over a time inter-
val of Tf =9 Myr the number of tidal disruptions is
Nd = TdT/ = 8.6, while the number of tidal captures is
smaller by a factor (r cri ,/a milI ) 3/2 ~ 0.08, Ni = YiT f ~ 0.69,
i.e: Ni/Nd — 0.08. The maximum distance from which stars
can originate is a max ~ 3.4 x lO 4 i?0. By using eq. 11311 . we
also find that the critical radius inside which the loss cone
should become empty is r cr it = 2 ■ 10 5 i?Q, slightly larger
than the inner end of the cusp.
From our simulations we find that Nd ~ 27, Ni = 4.1
and Ni/Nd = 0.15. Using only the inspirals which have sub-
Eddington heating rates gives Ni = 1.58 and Ni/Nd = 0.06
Hence the results for the analytical inspiral and disruption
rates Ni and Nd are within a factor of 3 to those found in
the simulations while the ratio Ni/Nd is correctly predicted.
0.001
0.0005
I
I !
( S
Tidal disruptions
Tidal disruptions
+ Inspirals
N„(a)
,N d (a) ~ a-°
ul I . . I I
1000 10" 10 5 10 s
Semi-major axis [R ]
10 7
10'
Figure 5. Initial semi-major axis distribution of tidally dis-
rupted stars (solid lines) and combined distribution of tidally dis-
rupted and captured stars (dotted). Shown arc differential rates
N(a) = (N(> a)—N(> a + Aa))/Aa. The combined distribution
has a maximum near the critical radius r cr it where the loss-cone
becomes empty, in agreement with theoretical predictions (Light-
man & Shapiro 1977). The combined distribution can be fitted
by two power-laws.
Although tidal disruptions are not the main focus of
the present paper, it is interesting to compare them with
analytic estimates since the physical processes responsible
for inspirals and disruptions are the same. The distribution
of semi-major axis of tidally d isrupte d stars has been calcu-
lated by Lightman & Shapiro [719771) . They found that the
peak in semi-major axis distribution should occur near r cr it
(sec cq. 1131 ). Inside r < r cr i t , the loss cone is empty, so stars
have to be scattered into the loss cone by two-body relax-
ation. In this case the number of disrupted stars should be
proportional to Nd(< r) ~ N(< r)/t r which leads to a scal-
ing Nd(< a) ~ a 3//2 for the number of encounters with semi-
major axis smaller than a (see eq. 1141 ). For the correspond-
ing differential function we find therefore Nd(a) ~ a 1 ' 2 . At
larger distance r > r cr it the loss cone is full ( "kick" regime) ;
the number of disrupted stars is then given by eq. |(T^J. This
leads to a scaling Nd(> a) ~ a~ , or, for the differential
function Nd(a) ~ a~ 2 .
Fig. shows the number of circularised and tidally dis-
rupted stars as a function of the initial semi-major axis. The
maximum occurs at a pca k — 2 • 10 4 Rq, somewhat smaller
than our analytical estimate for the critical radius rcrit -
For radii smaller than a pGa k, the combined distribution
of tidally heated and disrupted stars can be fitted by a rela-
tion N(a) ~ a 1,5 . This is somewhat steeper than predicted
by Lightman & Shapiro lfl977h . however the uncertainties in
the observed slope are quite large due to the small number
of events at small radii. If real, the reason for the discrep-
ancy with Lightman & Shapiro could be the capture of stars
by the IMBH followed by three body encounters with other
10 H. Baumgardt, C. Hopman and S. Portegies Zwart, J. Makino
Semi-major axis [R s ] M imbh [ m gJ
Figure 6. Fraction of tidally captured stars compared to all stars
ending in or near the IMBH as a function of the initial semi-major
axis a of the stars before tidal capture/disruption (see S(a) in eq.
[ED)- For small semi-major axis tidal capture dominates while
for larger radii tidal disruption becomes the dominant process
for feeding the central black hole since stars cannot spiral in any
more before being scattered away by other stars.
Figure 7. Semi-major axis of tidally circularised stars (dots)
compared to stars for which tidal heating does not lead to cir-
cularisation (crosses). Successful tidal inspiral is only possible for
stars on orbits with small semi-major axis. The threshold between
the two cases is in agreement with the prediction of this paper
and of Hopman et al. (2004) for the maximum inspiral distance
stars in the cusp. An inspection of the output data shows
that these processes contribute considerably to the tidal de-
struction rate but are not taken into account by our analytic
estimates. For radii larger than a pea k, we obtain a slow er de-
crease than predicted by Lightman & Shapiro 1(19771) . The
reason could be that the stars in the cusp drive the Brown-
ian motion of the IMBH. This influences the determination
of the semi-major axis of in-falling stars in case of large
semi-major axis.
Hopman & Alexander I (2005|) found that if stars move
slowly through phase space, the number of tidally captured
stars should be much larger than the number of tidally dis-
rupted stars for small radii, Ni(a) S> Nd(a); while for large
radii two-body relaxation becomes important and prevents
tidal inspiral so Nd(a) Ni(a). In Fig. |S|we compare the
number of stars that are tidally heated with all stars that
strongly interact with the IMBH. It can be seen that we
indeed find the predicted behaviour for S(a).
In §3.3 we demonstrated that the semi-major axis of a
stellar orbit has to be smaller than a max in order for the or-
bit of the star to be successfully circularised. Otherwise the
star is scattered away from its orbit due to encounters with
other stars. Fig.Qshows the initial semi-major axis of stars
which are tidally heated by the IMBH. Stars for which the
inspiral continues all the way until the orbit is circularised
are shown by filled circles while stars for which the heating
process is terminated since they are either scattered away or
experience super-Eddington heating rates during the inspi-
ral process are shown with crosses and triangles respectively.
Tidal capture is only successful if the initial semi-major
axis of the star a < 2 ■ lO 5 i? . The average inspiral distance
agrees quite well with the predicted value of a max . Stars cir-
cularising due to tidal heating fall into two categories: most
of them initially moved on orbits with e « 0.99 and come
from the inner cusp, while few came from relatively low-
eccentricity orbits with semi-major axis of a few 1000-Rq.
Most of the stars originating from low-a orbits experienced
dynamical interactions with previously inspiraled stars.
5.5 Further Evolution of captured Stars
In order to study the mass accretion rate and X-ray luminos-
ity of the IMBHs, we assume that only the innermost stars
are close enough to transfer a significant amount of mass to
the IMBHs and u se the model discussed by Kording, Falckc
& Markoff[(2002). They assume that X-rays are created in
accretion discs and jets and argue that a black hole with a
nearby companion is in the hard state if M > M C rit in which
case the disc luminosity is given by Lx = eMc 2 . At lower
accretion rates Lx = eM 2 c 2 /M CT i t . Following their paper,
we adopt e — 0.1 and Af cr it = O.1M0/Myr. The mass-loss
rates of the donor stars are direc tly tak en from the stellar-
evolution routines of Hurley et al. (2000). Stellar winds have
typical velocities of about 200 km/sec for LBVs and 1000
km/sec for Wolf-Rayet stars (Paumard et al. 2001), so de-
pending on the orbital separation of the innermost star and
the IMBH, gas lost by the star can also escape from the sys-
tem instead of ending up on the IMBH. A detailed study of
gas accretion would require SPH c alculations like the ones
presented by Cuadra et al. [(2005) , which are beyond the
scope of this paper. In order to account for the loss of gas,
we have plotted times when the escape velocity of the star
Tidal Capture of Stars by Intermediate- Mass Black Holes
11
42
i 1
1 1 1 1 1
2
s
u
ID
W
40
\
a
o .a
\
t-,
it >P
CO
_r
DO
f ^
MB2 UUP"! ' -
_0
38
36
iff
i
1 ,
I
i
-2 &
<
o
-4
4 6 8 10 12
T [Myrs]
Figure 8. Mass accretion rate and X-ray luminosity of the IMBH
of run 10. Crosses denote times when the distance of the star
closest to the IMBH is so large that most of the mass lost from
the star escapes from the star-IMBH system. Between 3 and 12
Myrs, several close companions to the IMBH formed through tidal
heating which transfer mass to the IMBH. For two events, the X-
ray luminosity of the IMBH reaches Lx = 10 40 erg/sec while
the stars are on the main-sequence, enough to explain the X-ray
luminosity of M82 ULX-1 and also within the right age range of
MGG-11.
is smaller than 200 km/sec with crosses. In these cases the
values for Lx should be regarded as upper limits.
Figs.|H|and|n|depict the X-ray luminosities of the IMBHs
as a function of time for runs 7 and 10. For both clusters, the
distance and nature of the innermost star changes rapidly
during the simulation as stars near the IMBH are turned into
compact remnants by stellar evolution and then scattered
out of the cusp by other stars or are tidally disrupted by the
IMBH. Run 10 experiences 4 prolonged ULX phases between
3 and 12 Myrs, each time due to massive main-sequence stars
brought close to the IMBH as a result of tidal heating. The
last two phases are within the right age limits for MGG-11
as determined by McCrady et al. and produce X-ray sources
with a luminosity exceeding 10 40 ergs/sec. Run 10 would
therefore be a good candidate to explain the nature of the
ULX in M82.
In run 7, three stars experience tidal inspiral between
3 < T < 4 Myrs and produce ULX sources before the stars
are tidally disrupted by the IMBH or turned into compact
remnants. At later times, the innermost stars are always too
far away from the IMBH for significant mass transfer, hence
run 7 is not able to explain the ULX in M82. In total we find
that in 10 of the 12 performed runs a ULX source is produced
at least once between 3 and 12 Myrs and that 4 out of the
12 runs create an X-ray source brighter than 2- 10 39 erg/sec
within the age range of MGG-11. The average time an X-
ray source with a luminosity brighter than 10 39 erg/sec is
present in our runs is 0.96 Myrs, while a source brighter than
10 40 erg/sec is present for 0.62 Myrs. These values would
be lower if tidal destruction of stars with super-Eddington
42
1
1
1 _
2
/ sec]
40
" f
9 -
s
\
o .a
no
u
_T
ate
MB2 ULX-1
ion r
log
38
<u
f
f
fH _
log Acc
36
Iflfl
.
* * -
i
-4
4 <
5 8
10
12
T [Myrs]
Figure 9. Same as Fig. [S] but now for run 7. This cluster also
experiences several inspiral events between 3 and 12 Myrs, but
none of them produces an X-ray source which is in the right age
range and bright enough to explain M82 ULX-1.
heating rates would be taken into account (see section Iq"31 .
however, at the moment our code does not follow the RLOF
phases which would significantly prolong the lifetime of the
ULX sources and increase their luminosities. We therefore
conclude that an IMBH in a dense star cluster like MGG-11
has a high chance of creating an ULX and that our model
can explain the ULX source in M82.
6 CONCLUSIONS
Observations of X-ray sources with luminosities higher than
the Eddington luminosity of stellar mass objects provide
strong support for the existence of IMBHs. However, the
mere existence of an IMBH does not guarantee the exis-
tence of ULXs; there has to be a mechanism which accounts
for the accretion of gas by the IMBH.
In this paper we confirm the hypothesis that massive
(M* > 1000 Mq) stars, which may be the progenitors of
IMBHs, form naturally in young dense clusters as a result
of many collisions between young stars. We assume that
the massive star indeed forms an IMBH, and we perform
N-body simulations of the interaction between the IMBH
and its host cluster, while accounting for tidal enc ounters
between stars and the IMBH. Madhusudhan et al. f2005)
found through stellar evolution studies that massive main-
sequence stars with masses M* > 8Mq in orbits of 6 to 30
times the radius of the donor star around an IMBH are nec-
essary to explain ULX sources. Our simulations show that
tidal heating of stars in young star clusters will create ex-
actly such systems. We find that, as a result of tidal heating,
stars are likely to circularise at distances of a few tidal radii
from the IMBH. Once the star has circularised, accretion of
stellar gas by the IMBH is sufficient to account for lumi-
nosities as high as L x Si 10 39 erg/s. On average, the X-ray
luminosities of the circularised stars exceed 10 39 ergs/sec for
12 H. Baumgardt, C. Hopman and S. Portegies Zwart, J. Makino
almost 10 6 yrs within the first 12 Myrs in our runs, adding
further credibility to the scenario that ULX are IMBHs ac-
creting gas lost from nearby companion stars.
Our conclusions differ from those reached by Blecha et
The main reason could be that the IMBHs in our
runs are more massive than the ones they considered and
they found that higher-mass IMBHs have a higher chance
of capturing companions. In addition, they also neglected
the effect of mass segregation, whereas in our runs we find
that mainly massive stars circularise around the IMBHs.
As we noted in §©, stars which are tidally captured by
an IMBH are typically among the most massive stars in the
cluster. The massive star loses much of its energy to other
stars by dynamical friction. In a single mass distribution
(as is often assumed in an analytical treatment of the prob-
lem^ diffusion in energy space is very slow (Bahcall & Wolf
1976), but when a spectrum of masses is present, energy loss
due to dynamical (non-elastic) interactions with field stars
(as opposed to tidal interactions with the IMBH) plays an
important role. This further emphasises the importance of
mass segregation.
In spite of the fact that some highly simplifying as-
sumptions were made, the final estimate for the rate result-
ing from our analytical treatment is of the same order of
magnitude as what we find from our simulations.
After the donor star leaves the main sequence, it forms
a compact remnant. For high-mass donor stars the remnant
will be a stellar mass BH and the subsequent evolution of the
IMBH-BH binary is driven by the emission of gravitational
waves until event ually th e binary members merge. Hopman
& Portegies Zwart l(2005l) have shown that the event rate for
this is likely to be high enough to be detectable by LISA, in
particular if the IMBH mass is larger than ~ 3 x 10 3 Mq.
Observations of gravitational waves from such a binary will
give further support for the scenario discussed in this paper.
In addition, if IMBHs are formed in star clusters near the
centres of galaxies and spiral in due to dynamical friction
on the field stars, mergers of the IMBHs with the MBHs
at galactic centres w ould a lso be important LISA sources
(Portegies Zwart et al l 2005F) .
The mechanism of stellar capture by gravitational wave
radiation is similar to that of tidal capture. Mass segregation
probably plays a less significant role in the case of the inspi-
ral due to gravitational wave emission by compact remnants
when spiralling in to a MBH of M. > 1O 6 M0, because in
that case the mass ratio of the field stars is less extreme than
is the case for a young dense stellar cluster. The number of
compact remnants in galactic centres is not known. Obser-
vationally the mass of these objects can be constrained in
our Galactic Centre by finding deviations from pure Kep-
lerian orbits, in particular pericenter-shift (Mouawad et al.
I 2004) . Current observational constraints are not conclusive,
but this situation will improve with more accurate measure-
ments of the stellar orbits in our Galactic Centre.
In addition to direct disruption and tidal inspiral, we
also find a large number of stars that experience one or sev-
eral strong tidal encounters, but are scattered to wider orbits
before they can circularise near the IMBH. As a result of the
strong tidal encounters, such tidally heated stars or "tidally
scattered" stars are expected to show signs of mixing, large
spin and stripping; this may be directly observable in the
Galactic centre (Alexander & Livi<E 2001). We found that on
average there are 45 stars which have one or more interac-
tions of less than 5rt with the MBH and are then scattered
away. We expect that this number scales linearly with peri-
center distance.
We have assumed that the IMBHs in the clusters formed
in a runaway merger of young stars, during a very early stage
of cluster evolution. However, other scenarios of IMBH for-
mation have been discussed in the literature. IMBHs may
form a s remnants of population III stars (e.g. Madau & Rees
I 20011) . or as the merger of stella r mass black holes due to
GW emission (Miller & Hamilto rTioOl]) . If IMBHs indeed
form in these scenarios, and they still reside in stellar clus-
ters, the stars in these clusters will have a much lower max-
imal mass than in the young clusters we discussed here. If
a star is tidally captured, accretion will most likely not lead
to high luminosity to account for the most luminous ULXs,
although accretion via the subsequent RLOF will lead to a
low-luminosity X-ray source.
ACKNOWLEDGEMENTS
We are gratefull to Sverre Aarseth for helping us with
NBODY4, and to Tal Alexander for interesting discussions.
This work is supported by Minerva grant 8484, the Dutch
Organization for Scientific Research (NWO under grant #
635.000.001), the Dutch Advanced School for Astronomy
(NOVA), and the Royal Netherlands Academy of Arts and
Sciences (KNAW). We thank the Sternwarte Bonn and the
University of Amsterdam for their hospitality.
REFERENCES
Aarseth, S. J., 1999, PASP, 111, 1333
Alexander, T., 1999, ApJ, 520, 137
Alexander & Livio, 2001, ApJ, 560, L143
Alexander, T., Hopman, C, 2003, ApJ, 590, L29
Alexander, T., Morris, M., 2003, ApJ, 590, L25
Bahcall, J. N., Wolf, R. A., 1976, ApJ, 209, 214
Bahcall, J. N., Wolf, R. A., 1977, ApJ, 216, 883
Baumgardt, H., Makino, J., Ebisuzaki, T., 2004a, ApJ, 613,
1133
Baumgardt, H., Makino, J., Ebisuzaki, T., 2004b, ApJ, 613,
1143
Blecha, L., et al., 2005, ApJ submitted, |astro-ph /0508597
Bond, J. R., Arnett, W. D., Carr, B. J., 1984, ApJ, 280,
825
Cohn, H., Kulsrud, R. M. 1978, ApJ, 226, 1087
Cuadra, J., Nayakshin, S., Springel, V., Di Matteo, T.,
2005, MNRAS in press, astro-ph/0502044
Fiorito, R., Titarchuk, L., 2004, ApJ, 614, L113
Frank, J., Rees, M. J., 1976, MNRAS, 176, 633
Freitag, M., 2001, A&A, 375, 711
Freitag, M., 2001, Class. Quantum Grav., 18, 4033
Freitag, M., 2003, ApJ, 583, L21
Freitag, M ., Gurkan, M. A., Rasio, F. A., 2005, MNRAS
submitted, astro-ph/0503130
Gutierrez, C. M., Lopez-Corredoira, M., 2005, ApJ, 622,
89
Hils, D., Bender, P. L., 1995, ApJ, 445, L7
Tidal Capture of Stars by Intermediate- Mass Black Holes 13
Hopman, C, Portegies Zwart, S.F., Alexander, T., 2004,
ApJ, 604, L101
Hopman, C, Alexander, T., 2005, ApJ, 629, 362
Hopman, C, & Portegies Zwart, S. F., 2005, MNRAS, 363,
L56
Hurley J. R., Pols O. R., Tout C. A., 2000, MNRAS 315,
543
Irwin, J. A., Bregman, J. N., & Athey, A. E., 2004, ApJ,
601, L143
Ishii, M., Ueno, M., Kato, M., 1999, PASJ, 51, 417
Ivanov, P. B., 2002, MNRAS, 336, 373, 2002
Kaaret, P., et al., 2001, MNRAS 321, L29
King A. R., Davies, M. B., Ward, M. J., Fabbiano, G.,
Elvis, M., 2001, ApJ, 552, 109
Kording, E., Falcke, H., Markoff, S., 2002, A&A, 382, L13
Kochanek, C. S., 1992, ApJ, 385, 604
Kuntz, K. D., Gruendl, R. A., Chu, Y-H, Chen, C.-H., Still,
M., Mukai, K., & Mushotzky, R. F., 2005, ApJ, 620, L31
Kustaanheimo, P., Stiefel, E. L., 1965, J. Reine Angew.
Math., 218, 204
Lee, H. M., Ostriker, J. P., 1986, ApJ, 310, 176
Li, Xiang-Dong, ApJ, 2004, 616, L119
Lightman, A. P., Shapiro, S. L., 1977, ApJ, 211, 244
Liu, J-F, Bregman, J. N., & Seitzer, P., 2004, ApJ, 602,
L249
Madhusudhan, N., et al., 2005, ApJ submitted,
|astro-ph/0511393|
Makino, J., Fukushige, T., Koga, M., & Namura, K., 2003,
PASJ, 55, 1163
McMillan, S. L. W., McDermott, P. N., Taam, R. E. 1987,
ApJ, 318, 261
Magorrian, J., Tremaine, S., 1999, MNRAS,309, 447
Miller, M. C. & Hamilton, D. P., 2001, MNRAS, 330, 232
McCrady, N., Gilbert, A. M., Graham, J. R., 2003, ApJ,
596, 240
Miller, J. M., Fabian, A. C, Miller, M. C, 2004, ApJ 614,
L117
Miller, M. C, & Colbert, E. J. M., 2004, International Jour-
nal of Modern Physics D., 13, 01
Miralda-Escude, J. & Gould, A., 2000, ApJ, 545, 847
Madau, P., & Rees, M. J., 2001, ApJ, 551, L27
Matsumoto, H., et al., 2001, ApJ, 547, L25
Mouawad, N., Eckart, A., Pfalzner, S., Schodel, R., Moul-
taka, J., Spurzem, R., 2004, Astronomische Nachrichten
Vol. 326, 2, 83-95
Peters, P. C, 1964, Physical Reviews B, 136, 1224
Podsiadlowski, Ph., 1996, MNRAS, 279, 1104
Portegies Zwart, S. F., Makino, J., McMillan, S. L. W., &
Hut, P., 1999, A&A, 348, 117
Portegies Zwart, S. F., McMillan, S. L. W., 2002 fillin
Portegies Zwart, S. F., Baumgardt, H., Makino, J., McMil-
lan, S. L., Hut, P., 2004, Nature, 428, 724
Portegies Zwart, S. F., Baumgardt, H., McMillan, S. L.,
Makino, J., Hut, P., Ebisuzaki, t., 2005, ApJ submitted,
|astro-ph/0511397|
Portegies Zwart. S. F.. Dewi, J., k Maccarone, T., 2004,
MNRAS, 355, 413
Portegies Zwart, S. F., Meinen, A. T., 1993, A&A, 280, 174
Press, W. H., Teukolsky, S. A., ApJ, 213, 183
Preto, M., Merritt, D., Spurzem, R., 2004, ApJ, 613, 109
Rappaport, S. A., Podsiadlowski, P., Pfahl, E., 2005, MN-
RAS, 356, 401
Ray, A., Kembhavi, A. K., Antia, H. M., 1987, A&A, 184,
164
Rees, M. J., 1988, Nature, 333, 523
Soria, R., Cropper, M., Motch, C, 2004, Chinese Journal
of Ast. & Astrophys. in press, astro-ph/040913
Sigurdsson, S., Rees, M. J., 1997, MNRAS 284, 318
Stohmayer, T. E., Mushotzky, R. F., 2003, ApJ, 586, 61
Swartz, D. A., Ghosh, K. K., Tennant, A. F., & Wu, K.,
2004, ApJ, 154, S519
Syer, D., Ulmer, A., 1999, MNRAS, 306, 35
Wang, J., Merritt, D., 2004, ApJ, 600, 149
This paper has been typeset from a T^X/ F/IgX file prepared
by the author.
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11